Population stratification and admixture are critical problems in case-control association studies. Spurious associations can arise when mixed populations, with different allele frequencies, are used. The AA population is a recently admixed population with estimated mean European ancestry proportion of 17-20% [e.g. (Guthery et al., 2007, Reiner et al., 2005). In contrast, in this study population, there was an average of 11% European ancestry admixture. To minimize the effect of admixture we applied two independent methods and excluded population stratification between cases and controls. The AIMs data also validated the accuracy of the three generations ethnic self-identification in our dataset. The few exceptions (~1%) may be due to the individuals' inaccurate knowledge of their own family histories.
